Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <CAFcO6XN7JDM4xSXGhtusQfS2mSBcx50VJKwQpCq=WeLt57aaZA@mail.gmail.com>
 <87edwrl9i8.wl-tiwai@suse.de>
In-Reply-To: <87edwrl9i8.wl-tiwai@suse.de>
From:   butt3rflyh4ck <butterflyhuangxx@gmail.com>
Date:   Mon, 5 Sep 2022 01:06:05 +0800
Message-ID: <CAFcO6XPxCN7DEXbUvtRTKq+ku0OJA21gCz_dqMacDpsNUrsUhg@mail.gmail.com>
Subject: Re: A new null-ptr-deref Write bug in snd_pcm_format_set_silence
To:     Takashi Iwai <tiwai@suse.de>
Cc:     perex@perex.cz, tiwai@suse.com, cuibixuan@linux.alibaba.com,
        alsa-devel@alsa-project.org, LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

I have tested this patch, it works well.

Regards,
 butt3rflyh4ck.

On Sun, Sep 4, 2022 at 6:27 PM Takashi Iwai <tiwai@suse.de> wrote:
>
> On Sun, 04 Sep 2022 11:48:37 +0200,
> butt3rflyh4ck wrote:
> >
> > Hi, there is a new null-ptr-deref Write bug in
> > snd_pcm_format_set_slience in sound/core/pcm_misc.c in the latest
> > upstream kernel and can reproduce it.
> > We call SNDCTL_DSP_SYNC and SNDCTL_DSP_SPEED in multiple threads to
> > trigger the vulnerability.
> >
> > See the Call Trace:
> > ==================================================================
> > Call Trace:
> >  <TASK>
> >  __dump_stack lib/dump_stack.c:88 [inline]
> >  dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
> >  kasan_report+0xb1/0x1e0 mm/kasan/report.c:495
> >  check_region_inline mm/kasan/generic.c:183 [inline]
> >  kasan_check_range+0x13d/0x180 mm/kasan/generic.c:189
> >  memset+0x20/0x40 mm/kasan/shadow.c:44
> >  snd_pcm_format_set_silence sound/core/pcm_misc.c:441 [inline]
> >  snd_pcm_format_set_silence+0x215/0x350 sound/core/pcm_misc.c:424
> >  snd_pcm_oss_sync+0x60e/0x800 sound/core/oss/pcm_oss.c:1690
> >  snd_pcm_oss_ioctl+0x2087/0x3420 sound/core/oss/pcm_oss.c:2634
> >  vfs_ioctl fs/ioctl.c:51 [inline]
> >  __do_sys_ioctl fs/ioctl.c:870 [inline]
> >  __se_sys_ioctl fs/ioctl.c:856 [inline]
> >  __x64_sys_ioctl+0x193/0x200 fs/ioctl.c:856
> >  do_syscall_x64 arch/x86/entry/common.c:50 [inline]
> >  do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
> >  entry_SYSCALL_64_after_hwframe+0x63/0xcd
> > ==================================================================
> > We can see the function snd_pcm_format_set_silence code below:
> > ```
> > int snd_pcm_format_set_silence(snd_pcm_format_t format, void *data,
> > unsigned int samples)
> > {
> >         int width;
> >         unsigned char *dst;
> >         const unsigned char *pat;
> >
> >         if (!valid_format(format))
> >                 return -EINVAL;
> >         if (samples == 0)
> >                 return 0;
> >         width = pcm_formats[(INT)format].phys; /* physical width */
> >         pat = pcm_formats[(INT)format].silence;
> >         if (!width || !pat)
> >                 return -EINVAL;
> >         /* signed or 1 byte data */
> >         if (pcm_formats[(INT)format].signd == 1 || width <= 8) {
> >                 unsigned int bytes = samples * width / 8;
> >                 memset(data, *pat, bytes);    ///// [1] ---------> data is NULL
> >                 return 0;
> >         }
> >        ......
> > }
> > ```
> > [1], the data pointer is NULL, we can know snd_pcm_format_set_silence
> > called in line 1690 in sound/core/oss/pcm_oss.c from call stack trace.
> > let we see code below:
> > ```
> > static int snd_pcm_oss_sync(struct snd_pcm_oss_file *pcm_oss_file)
> > {
> >         int err = 0;
> >         unsigned int saved_f_flags;
> >         struct snd_pcm_substream *substream;
> >         struct snd_pcm_runtime *runtime;
> >         snd_pcm_format_t format;
> >         unsigned long width;
> >         size_t size;
> >
> >         substream = pcm_oss_file->streams[SNDRV_PCM_STREAM_PLAYBACK];
> >         if (substream != NULL) {
> >                 runtime = substream->runtime;
> >                 if (atomic_read(&substream->mmap_count))
> >                         goto __direct;
> >                 err = snd_pcm_oss_make_ready(substream);
> >                 if (err < 0)
> >                         return err;
> >                 atomic_inc(&runtime->oss.rw_ref);
> >                 if (mutex_lock_interruptible(&runtime->oss.params_lock)) {
> >                         atomic_dec(&runtime->oss.rw_ref);
> >                         return -ERESTARTSYS;
> >                 }
> >                 format = snd_pcm_oss_format_from(runtime->oss.format);
> >                 width = snd_pcm_format_physical_width(format);
> >                 if (runtime->oss.buffer_used > 0) {
> > #ifdef OSS_DEBUG
> >                         pcm_dbg(substream->pcm, "sync: buffer_used\n");
> > #endif
> >                         size = (8 * (runtime->oss.period_bytes -
> > runtime->oss.buffer_used) + 7) / width;
> >                         snd_pcm_format_set_silence(format,
> >                                                    runtime->oss.buffer
> > + runtime->oss.buffer_used,   ///// [2]
> >                                                    size);
> >                         err = snd_pcm_oss_sync1(substream,
> > runtime->oss.period_bytes);
> >                         if (err < 0)
> >                                 goto unlock;
> >                 } else if (runtime->oss.period_ptr > 0) {
> >
> > ```
> > [2]  runtime->oss.buffer + runtime->oss.buffer_used is the data
> > pointer, but runtime->oss.buffer is NULL here but it doesn't make
> > sense.
> > runtime->oss.buffter is allocated by kvzalloc, if runtime->oss_buffer
> > is NULL, it would return an ENOMEM error.
> > Maybe I think there is a race condition, the runtime->oss.buffer is
> > freed and set to NULL but we can use runtime->oss.buffter via ioctl.
>
> Yeah, likely it's in a small race window between two calls.
>
> Could you try the patch below?
>
>
> thanks,
>
> Takashi
>
> ---
> --- a/sound/core/oss/pcm_oss.c
> +++ b/sound/core/oss/pcm_oss.c
> @@ -1672,14 +1672,14 @@ static int snd_pcm_oss_sync(struct snd_pcm_oss_file *pcm_oss_file)
>                 runtime = substream->runtime;
>                 if (atomic_read(&substream->mmap_count))
>                         goto __direct;
> -               err = snd_pcm_oss_make_ready(substream);
> -               if (err < 0)
> -                       return err;
>                 atomic_inc(&runtime->oss.rw_ref);
>                 if (mutex_lock_interruptible(&runtime->oss.params_lock)) {
>                         atomic_dec(&runtime->oss.rw_ref);
>                         return -ERESTARTSYS;
>                 }
> +               err = snd_pcm_oss_make_ready_locked(substream);
> +               if (err < 0)
> +                       goto unlock;
>                 format = snd_pcm_oss_format_from(runtime->oss.format);
>                 width = snd_pcm_format_physical_width(format);
>                 if (runtime->oss.buffer_used > 0) {


-- 
Active Defense Lab of Venustech